Search

• Substantial surveys identify that bus queuing, merging, and repeated opening doors can significantly affect bus dwell time. • A Markov chain is designed to characterize the effect of vehicles in adjacent lane and passengers on bus merging. • A comprehensive model is proposed to estimate bus dwell time and bus queuing time. • The actual performance of the comprehensive model is validated in real cases in three cities. • The probability of bus overtaking is highly correlated with bus service time and average stop service time. The precise estimation of bus dwell time (BDT) is immensely important as it directly impacts the reliability of bus schedules and the accuracy of arrival time predictions. Various factors—such as the behavior of buses merging into the main traffic flow, uneven distribution of passengers on board, and frequent bus queues—jointly lead to a high variability of the BDT. To achieve an accurate estimation of BDT across various stop types, a comprehensive BDT model is developed in this paper. First, based on berth occupancy, probabilistic models are constructed for different bus arrival scenarios, which are then used to study corresponding bus queuing times. Second, the Bureau of Public Roads function is employed to characterize the obstruction effect occupants on the bus impose on boarding passengers. Combined with this function, a bus service time model is proposed considering the bus opening doors more than once and in-vehicle crowding. Third, the bus merging process is discretized to cope with the volatility of the time interval required for a bus to leave the bus bay. A Markov chain model is designed to calculate the time buses have to wait before merging. Case studies conducted in Qingdao, Chongqing and Beijing, China, demonstrate the effectiveness of the comprehensive model both during off-peak and peak periods. Furthermore, this paper incorporates overtaking rules into the comprehensive model to examine the influence of bus overtaking willingness and various overtaking policies on bus queuing times. Through numerical analysis, the key link is between willingness to overtake, service time and the likelihood of bus overtaking is disclosed. [ABSTRACT FROM AUTHOR]

The transition to alternative energy sources and the adoption of on-demand operating modes in urban bus systems are crucial steps towards reducing carbon footprints and improving public transit services. This paper presents a two-phase approach for the collaborative optimization of charging schedules and passenger services, aimed at enhancing the operation of on-demand electric bus systems. First, we propose a label-setting dynamic programming algorithm that enables the efficient generation of bus-trips for each bus line in response to passenger requests. Second, we introduce a time–space network optimization model that facilitates integrated multiple bus-trip planning for the transit network, involving multiple bus lines and charging spots. The model selects bus-trips from various time–space arcs, which represent passenger carrying, bus deployment, and bus charging activities. To validate the effectiveness of our approach, we conduct a case study using real-world data from bus lines in Beijing, China. Computational results demonstrate that our approach can handle on-demand electric bus operations within minutes of solution time, efficiently serving over 2,000 passengers. Practically, our approach achieves a notable reduction in average transit time and effectively reduces the waste of public transit resources. The proposed approach can serve as a beneficial tool for decision-makers and operators seeking to enhance the performance and environmental impact of their electric bus systems. • This paper aims at enhancing on-demand bus operational efficiency. • This paper proposes an optimization approach for electric bus operations. • This paper provides practical solutions for responsive transit systems. [ABSTRACT FROM AUTHOR]